The Sports PageTwo Hundred Seventy Division Champions. This Year's Will Be the Worst of Them.
The question we ask every Sunday: what did we claim this week, and did reality agree?
The scoreboard. Five issues, one gradeable prediction: one hit, zero missed, zero still pending from this week. Four pieces made no forward forecast at all — they were methods and retrospective analysis, and that is legitimate work. But the one piece that did make a prediction made a big one.
Here is the surprising part. The most interesting thing this week is not a miss. It is a hit that was hiding in plain sight: on Friday we published a 70% probability that the AL West division champion would finish at .500 or below for the first time in 57 years. Saturday’s results removed every scenario where that prediction fails, so the 70% is now 100%. Read the wording carefully though: at or below .500 is settled; below .500 is not, and will not be until tonight.
Why we bother: a prediction nobody grades is just a story.
The Week in Review: Issues #178–182
- #178Sep 22 — The JUCO Detour: College Baseball’s Portal Workaround Is Already Here — Junior college players aren’t transferring — they’re arriving. Brady Ballinger hit .353 at Kansas after .433 at CSN in Las Vegas. Smart programs are already loading rosters through the JUCO pipeline the EO never touched.
- #179Sep 23 — Your Team Held the Ball Longer. That Just Means They Were Ahead. — Time of possession correlates with a team’s second-half win rate at +0.216. Control for the score and it falls to −0.003. Across 5,278 team-games, possession time tells you the score, not the team.
- #180Sep 24 — The Crowd Splits Every Game in Half. You Still Lose a Thousand Dollars. — A point spread is a fair estimate of the outcome. The price is not. Both sides at −110 add to 104.76%, and that 4.76% surplus never plays a down. Break-even requires being right 52.4% of the time.
- #181Sep 25 — Turnovers Win Games. They Do Not Win Seasons. — A season’s turnover margin explains 3% of the next season’s. Yards per play explains 15.4%. Turnovers feel like character and behave like weather. Across 288 team-seasons, a team that wins the turnover battle one year wins it again the next only slightly more than randomly.
- #182Sep 26 — Two Teams Are Tied With Losing Records. One Is Going to the Playoffs. — Houston and Texas both 79–80 with three games left. Across 270 division winners since 1969, not one finished below .500. We gave a 70% probability that one of them would. They both won Saturday.
The Call That Came True: Issue #182
What the Math Said. What Happened.
On Friday, we built a simulation of the remaining schedule from 79–80 with three games each. Houston was facing the Athletics (63–96), Texas was facing the Twins (74–85). We gave Houston a 58% chance per game, Texas a 51% chance. You can read the full piece at Issue #182. Here is what the distribution said:
The 30% path — the only historical precedent — required both teams to sweep their remaining three games. Both teams won on Saturday, moving to 80–81. The sweep path is now closed. The champion cannot exceed 81–81.
We said the probability of an unprecedented finish was 70%. It is now 100%. The winner today will be either the first .500 division champion ever, or the first below-.500 one. Both are new. Check the standings; the game starts at 2pm ET.
The Full Prediction Scorecard
| # | Prediction | Outcome so far | Grade |
|---|---|---|---|
| 182 | 70% probability the AL West champion finishes at .500 or below — first time in 57 years, 270 division winners | Both teams went to 80–81 Saturday. Champion cannot exceed 81–81. The 30% historical-precedent path is eliminated. | HIT — confirmed before the final game |
| 181 | No forward prediction. Turnover persistence is a measurement (3.0% year-over-year), not a forecast. | — | N/A — method piece |
| 180 | No forward prediction. The vig is a structural fact, not a claim about future outcomes. | — | N/A — numeracy piece |
| 179 | No forward prediction. Possession collapse is a confound measurement. | — | N/A — method piece |
| 178 | No forward prediction. JUCO analysis is retrospective and structural. | — | N/A — EO series |
| 175 | Teams winning by 20+ points beat the spread 47.9% the following week — a directional lean against blowout winners | Penn State won 55–13 over Buffalo in Week 3; they lost outright to Wisconsin 24–20 in Week 4. One clean data point in the predicted direction. Insufficient to confirm or deny across the full population for one week. | PARTIALLY HIT — one data point, right direction |
| 165 | Week 1 results explain ~21% of final season outcome; Notre Dame and Oregon dominant openers signal genuine quality | Notre Dame 4–0, beat Purdue 49–10. Oregon 3–1 — the one loss to Oklahoma State in Week 2 is the complication; beat #12 USC on the road in Week 4. | PENDING — Notre Dame: tracking HIT. Oregon: 3-1 complicates the clean story. |
| 143 | South Carolina projected at 7.8 wins out of 12 (preseason model; they were ranked 35th) | Now 2–2 overall, 0–2 SEC after losses to Mississippi State and Alabama (49–18). Need to go 5–3 in remaining eight games to hit projection. | PENDING — yellow flag. The model needs more from the next eight games than it is currently showing. |
| 168 | Detroit’s 14–27 one-run record is luck; year-over-year correlation near zero across the league. 2027 Tigers will not repeat it. | Tigers finished with a losing record despite outscoring opponents for the year. Pattern matches prediction exactly. The real grade is April 2027. | PENDING — 2027 check |
What We Got Right
Correction — 27 September 2026, 6am
This edition first went out headlined “Not One Finished Below .500. Now One Does.” That was wrong, and it was wrong in the way we have a rule against: it asserted something that had not happened yet.
Houston and Texas are both 80–81 with one game each still to play. The champion finishes 81–81 if either wins, and 80–82 only if both lose. So the chance of a champion finishing strictly below .500 is about 21%, not 100%. What is locked in is that the winner finishes at or below .500, and that it will be the worst division champion in 57 years — the previous worst was 82–80.
The body of this piece said “at .500 or below” correctly throughout. The headline did not, and the headline is what most people read. Corrected within the hour; the original wording is quoted above so the change is visible rather than silent.
The AL West Call
Issue #182 on Friday was the paper’s most precise probabilistic forecast since the Soto injury call in the spring. We gave a 70% probability of something that had never happened in 57 years of baseball. Both teams then won on Saturday, which removed the only path to a historical outcome and locked in the unprecedented one. The forecast was right and it was right for the right reason: we did not say “one of these teams is bad”. We computed from game odds and remaining schedule that the 82–80 finish required a sweep by teams averaging less than 50% across the season. The math did the work.
Notice also what the model said was most likely: 81–81 at 52%. That is now the ceiling. The single most probable outcome in Friday’s simulation is now also the best possible outcome. When a model’s most likely scenario becomes the ceiling of reality, that is a sign the model understood the situation. You can hold onto that: when the most likely bar becomes the upper bound, you were not lucky. You were right about where the probability lived.
What We Got Wrong
South Carolina Is Trending the Wrong Way
Nothing from this week’s issues was materially wrong. The methods pieces hold. The JUCO piece is observational and accurate. The big prediction hit.
But the carry-over from Issue #143 — our projection that South Carolina would win 7.8 games out of 12 — is showing early strain. They are 2–2 and 0–2 in SEC play after a 49–18 loss to Alabama. That is not a fatal result for a 7.8-win projection; they have eight games left and need to go 5–3. But 0–2 in conference play against Mississippi State and Alabama is not the opening the model expected. Notice what the loss to Alabama tells you: a 31-point margin against a team you had projected to beat roughly half your conference games is meaningful evidence. We are not calling a miss yet. We are watching, and saying out loud that the projection has harder work to do now.
“A 70% probability confirmed at 100% is not luck. It is the model earning its keep.”
— The Sports Page, Sunday Edition No. 025What Yesterday’s College Football Showed
Week 4 of the college football season ran Saturday, and the numbers carried two stories worth naming.
The first is Penn State. They beat Buffalo 55–13 last week. That is a 42-point margin, a result that should make Penn State a comfortable home favorite this Saturday. Instead, Wisconsin came to State College and won 24–20. Penn State lost outright as what was almost certainly a double-digit home favorite. Issue #175, published September 19, made a directional claim: teams winning by 20 or more points cover the spread only 47.9% the following week. Penn State is the cleanest possible data point for that thesis. One game is not a pattern. But one game is not nothing either, especially when the result is an outright home loss.
The second story is Texas A&M. They went to LSU and lost 35–6. Last week, you will remember, Kentucky beat them at home as a 16.5-point underdog — the largest upset of Week 3 by raw probability. The market repriced A&M immediately; they were an 8.5-point road underdog at LSU rather than a double-digit favorite. Then they lost by 29. Texas A&M is now 1–3 on the season. The market was right to downgrade them. An upset is always explained afterward. But sometimes the explanation is simply: the team is not good.
The Improbability Lesson
Notice the difference between those two stories. Penn State being upset by Wisconsin is evidence for the blowout-winner regression thesis, but it is still noise in a one-week window. Texas A&M being outscored 35–6 in the game after an 18-point underdog beat them is something different. That is a body of evidence accumulating.
The lesson both carry is the same one the AL West is teaching today: “nobody saw this coming” almost always means “nobody who was paying attention to the right numbers.” A model that gave Kentucky 12% against A&M was not surprised that the upset happened. It was only uncertain about which game would produce it. Across a 49-game Saturday card, roughly five or six underdogs should win on a normal week. The list of 1-in-9 shots is long. Kentucky happened to be on it. Your job — and this paper’s job — is to treat a 1-in-9 result as exactly that: evidence the model was used correctly, not evidence it was wrong.
Notre Dame remained steady: 4–0, beat Purdue 49–10. Oregon beat #12 USC on the road, 41–27, with their backup quarterback. The teams that looked like the good teams in Week 1 are, with one notable exception (Oregon’s Week 2 stumble against Oklahoma State), still the good teams. That is what 21% predictive power from Week 1 looks like in practice: not certainty, but a prior that updates more slowly than the market does.
What to Watch Today, and Why
The NFL card is ranked below by closeness to a coin flip — not by market size, not by fame. The game nearest to an even split is the game where the result carries the most information, because that is where the most playoff probability changes hands per play. We hold no NFL forecast of our own yet. Four weeks is not a model. The probabilities below are from the closing spread, with a 13.5-point standard deviation.
The game closest to a coin flip is the Rams at Denver (Rams −2.5, Sunday Night Football, 8:20pm ET). But the most interesting game on the card by championship leverage is the Raiders at New Orleans (Saints −3, 4:25pm ET). The Raiders are 2–0. They beat the Dolphins by 14 and the Chargers by 12. They lead the NFL in sacks. The market is giving them away as a three-point road underdog because they are the Raiders and the Saints are at home. A team that has already beaten two teams who are favorites today is getting less respect than a “we will see” dismissal. That is exactly the kind of structural mismatch the vig piece was about.
| Game | Line | Fav wins | Kickoff (ET) |
|---|---|---|---|
| LAR at DEN 🏈 SNF | LAR by 2.5 | 55% | 8:20 PM NBC |
| LV at NO — Raiders 2–0 | NO by 3.0 | 59% | 4:25 PM CBS |
| CIN at PIT | CIN by 3.5 | 60% | 1:00 PM CBS |
| BAL at DAL | BAL by 3.5 | 60% | 4:25 PM CBS |
| NYJ at DET | DET by 6.5 | 69% | 12:00 PM FOX |
| LAC at BUF | BUF by 7.0 | 70% | 1:00 PM FOX |
Probabilities from closing spread, 13.5-point standard deviation. Highlighted rows are within one score of a coin flip. We hold no NFL forecast of our own yet — three weeks is not a model.
Over-Reactions and Under-Reactions
Over-Reactions
The AL West champion is not a bad team that snuck through. They are a team that played in a weak division in a year when both the Astros and Rangers had meaningful injuries and roster turnover. Do not write the AL West off as broken. Write it as five teams that were bad at approximately the same level. The trophy still goes to whoever wins today. One bad division in one season is not a structural crisis.
Penn State is not in freefall. They lost one game at home, as a big favorite, to a Wisconsin team that has beaten teams like this before. One result is one result. Watch the next three before calling it a collapse.
Under-Reactions
Texas A&M at 1–3 is worth taking seriously. This is not a team having bad luck. They lost to Kentucky at home as a 16.5-point favorite, then lost to LSU by 29 on the road. The market repriced them aggressively and the market was right. Do not call this variance. Call it what the 35–6 score suggests: a team with real problems.
South Carolina at 0–2 in SEC play deserves attention. A 7.8-win projection assumes they beat the teams they should beat in conference play. A 31-point loss to Alabama is not “just the schedule.” The next game tells us more.
What Readers Read · Sep 20–Sep 26
Readership data pending — the analytics API was unreachable at publish time. The archive is at thesportspage.net.
What to Take Home, and the Road Ahead
Take this home: the AL West story ends today, and you will know the exact record by dinner. Watch whether it is 81–81 or something worse. Either way, you are watching history. You knew it was coming before anyone else did, and you knew why. That is the whole point of keeping a scorecard.
This week’s queue has pieces on cost of fandom, lottery pricing, and the economics of what actually moves standings. The Mets playoff picture is still alive; the baseball postseason begins Monday, September 28. Our log-odds projections for each series, made before they start and graded as they go, will appear as each matchup is set. That is the kind of accountability the Sunday Edition exists for.
South Carolina plays again this week. Watch carefully. The Issue #143 prediction needs a win to stay plausible.
“Twenty-five Sundays in. One hit that mattered. Zero misses to confess from this week. The math held.”
— The Sports Page, on the long game
